Tutorial on multivariate autoregressive modelling.

Area of Science:

• Biomedical Engineering
• Signal Processing
• Systems Biology

Background:

• Multivariate autoregressive (MAR) modeling is essential for studying linear relationships between multiple time-series signals.
• Applications in biomedical engineering include analysis of cardiovascular dynamics and electroencephalographic (EEG) signals.
• MAR modeling facilitates the determination of physiologically relevant connections between measured biological signals.

Purpose of the Study:

• To explain the theoretical background of multivariate autoregressive (MAR) modeling.
• To illustrate the application of MAR modeling in analyzing physiological signals.
• To demonstrate how MAR models can describe causality, delays, and closed-loop effects.

Main Methods:

• The paper details the theoretical framework of MAR modeling.
• MAR models predict each variable's value based on its past values and those of all other time series.
• Model order, representing the number of past values used, is a key parameter.

Main Results:

• MAR models can effectively describe complex inter-signal dynamics, including causality and delays.
• The study illustrates MAR modeling using a practical example involving systolic blood pressure, RR interval, and lung volume.
• This approach provides insights into the linear relationships and physiological connections within biological systems.

Conclusions:

• MAR modeling is a powerful tool for analyzing complex physiological signals in biomedical engineering.
• It enables the identification of physiologically relevant connections, causality, and dynamic interactions.
• The illustrated model demonstrates the utility of MAR for understanding cardiovascular and respiratory interactions.